JP5629481B2 - Damage diagnosis system - Google Patents

Description

The present invention also relates to damage diagnosis system utilizing the Lamb wave.

For example, in fields where both strength and weight reduction for materials such as aircrafts are required, it is indispensable to apply a composite material such as CFRP in order to meet such demands. In addition, damage detection technology (health monitoring technology) is attracting attention in order to ensure high reliability and more efficient design of the composite material structure.
Patent Documents 1 and 2 describe a damage detection apparatus using an FBG (Fiber Bragg Grating) optical fiber sensor as an apparatus for detecting damage, defects, and the like of such a composite material. Optical fibers have recently been reduced in diameter (for example, 52 [μm] in diameter), and even if they are embedded in a structure, the strength of the structure does not decrease so much. Has advantages.

  According to the inventions described in Patent Documents 1 and 2, the structural composite material is configured between the piezoelectric element fixedly disposed at a predetermined position of the structural composite material, the conductive wire that transmits a signal to the piezoelectric element, and the piezoelectric element. An optical fiber sensor having a grating part that is fixedly disposed across a composite material that reflects light of a predetermined wavelength on the core part, a light source that irradiates the core part with light, and a characteristic that detects characteristics of reflected light from the grating part The detecting means is used to detect the damage from the change in the output of the characteristic detecting means by exciting the piezoelectric element. As the characteristic detection means, a spectrum analyzer that detects the frequency characteristic of reflected light from the grating section is used.

  Furthermore, in the invention described in Patent Document 1, comparison is made with detection data obtained from a normal structural composite material obtained in advance, or as another method, from when no specific vibration is caused by a frequency distribution detected by a spectrum analyzer. It is described that a threshold may be set for the fluctuation value of, and if it is less than that, it may be determined that there is damage (paragraph 0032).

  In Patent Document 2, a spectrum analyzer is provided with two optical filters, and the spectrum analyzer outputs reflected light to the arithmetic processing device through the two optical filters, thereby detecting the wavelength vibration signal of the reflected light with high sensitivity. It has been proposed that the arithmetic processing unit calculates a value (DI value) corresponding to the magnitude of damage to the subject based on the obtained wavelength vibration signal.

As a technique for detecting damage, research has been conducted to diagnose the occurrence of damage from changes in the received waveform by transmitting and receiving ultrasonic waves in the form of Lamb waves. A Lamb wave is an ultrasonic wave propagating through a thin plate, and has a relatively small attenuation and propagates through the plate structure over a long distance. Therefore, the Lamb wave is an ultrasonic wave propagation form suitable for damage detection.
Lamb waves have two characteristics, multimodality and velocity dispersion (frequency dependency), and there are a plurality of modes with different velocities depending on the plate thickness and frequency. Due to this complicated feature, conventionally, damage detection has been performed using only information on a specific frequency of Lamb waves.

JP-A-2005-98921 JP 2007-232371 A US Pat. No. 5,493,390

The present invention makes it possible to measure the modal dispersibility over a wide band by utilizing this characteristic of the dispersibility of Lamb waves, and more information useful for damage detection than before can be obtained, and the peel length can be quantitatively evaluated. , detection of damage highly reliable with high precision, and to provide a damage diagnosis system which enables diagnosis.

The invention according to claim 1 for solving the above-described problems includes an excitation device that applies ultrasonic vibration of a broadband Lamb wave to a subject,
A vibration detection sensor for detecting a broadband Lamb wave transmitted from the subject;
A processing device that controls the oscillation operation of the vibration exciting device, processes the output value of the vibration detection sensor, and outputs a measurement result;
The processor is
By converting the output value of the vibration detection sensor obtained at the time of vibration by the vibration device to obtain the propagation intensity distribution data developed in two dimensions of frequency and propagation time, the asymmetric mode of Lamb wave is identified. For one or more specific asymmetric modes selected from the identified Lamb wave asymmetric modes, the propagation time at which the maximum value of the propagation intensity occurs for each frequency is calculated, and the frequency and propagation at which the maximum value occurs. Identify time relationships,
In contrast to the relationship between the identified frequency and propagation time, and the relationship between the frequency and propagation time specified in advance in a subject having a known damage state, which is stored in the storage unit, the damage scale for the subject is estimated. It is a damage diagnosis system that performs calculation and displays.

In the invention according to claim 2, the excitation device and the vibration detection sensor are respectively installed at the same position on the front and back of the subject,
The processing device oscillates asymmetrically by oscillating anti-phase waves in the vibration devices on the front and back sides, converts the calculated data by adding the output values of the vibration detection sensors on the front and back sides, and converts the frequency and The damage diagnosis system according to claim 1, wherein the propagation intensity distribution data developed in two dimensions of propagation time is obtained .

According to a third aspect of the present invention, the processing apparatus includes:
Instead of specifying the relationship between the frequency and propagation time at which the maximum value of the propagation intensity distribution data occurs,
Calculate the propagation time at which the maximum value of the propagation intensity occurs for each frequency and plot the data indicating the relationship between the specified frequency and the propagation time on a graph in which the horizontal axis represents the frequency and the vertical axis represents the propagation time. Calculate the slope of the data in the frequency range as the rate of change,
The calculated change rate is compared with the change rate calculated in advance in a subject whose damage state is known, which is accumulated in a storage unit, and is displayed by performing an estimation calculation of a damage scale for the subject. The damage diagnosis system according to claim 1 or claim 2.

According to a fourth aspect of the present invention, the processing apparatus includes:
Instead of calculating the rate of change,
Calculate the amount of decrease in the propagation time at the maximum value of the propagation intensity distribution data with respect to the case where the subject is not damaged,
The damage diagnosis system according to claim 3 , wherein an estimation calculation of a damage scale for the subject is performed and displayed based on the calculated decrease amount .

The invention according to claim 5 is an excitation device that applies ultrasonic vibration of a broadband Lamb wave to a subject;
A vibration detection sensor for detecting a broadband Lamb wave transmitted from the subject;
A processing device that controls the oscillation operation of the vibration exciting device, processes the output value of the vibration detection sensor, and outputs a measurement result;
The processor is
By converting the output value of the vibration detection sensor obtained at the time of vibration by the vibration device, the propagation intensity distribution data expanded in two dimensions of frequency and propagation time is obtained, and the symmetry mode of the Lamb wave is identified. For one or more specific symmetric modes selected from the symmetric modes of the identified Lamb wave, the propagation time at which the maximum value of propagation intensity occurs for each frequency is calculated, and the frequency and propagation at which the maximum value occurs. Identify time relationships,
In contrast to the relationship between the identified frequency and propagation time, and the relationship between the frequency and propagation time specified in advance in a subject having a known damage state, which is stored in the storage unit, the damage scale for the subject is estimated. It is a damage diagnosis system that performs calculation and displays.

In the invention according to claim 6, the vibration device and the vibration detection sensor are respectively installed at the same position on the front and back of the subject,
The processing device oscillates symmetrically by causing the vibration devices on the front and back to oscillate in-phase waves, converts the calculated data by adding the output values of the vibration detection sensors on the front and back, and converts the frequency and 6. The damage diagnosis system according to claim 5, wherein propagation intensity distribution data developed in two dimensions of propagation time is obtained .

According to a seventh aspect of the present invention, the processing apparatus includes:
Instead of specifying the relationship between the frequency and propagation time at which the maximum value of the propagation intensity distribution data occurs,
Calculating the amount of increase in the propagation time at the maximum value of the propagation intensity distribution data relative to the case where the subject is not damaged;
The damage diagnosis system according to claim 5 or 6 , wherein an estimation calculation of a damage scale for the subject is performed and displayed based on the calculated increase amount .

The invention according to claim 8 is characterized in that the processing device displays a relationship between the frequency and the propagation time in the specified asymmetric mode or symmetric mode instead of displaying the damage scale for the subject obtained by performing the estimation calculation. The damage diagnosis system according to any one of claims 1 to 7 , wherein the calculated change rate of the propagation time, the calculated decrease amount, or the calculated increase amount is displayed .

Lamb waves have a symmetric mode (S mode) having a symmetric displacement with respect to the thickness center of the plate-like vibration propagation medium, and an asymmetric mode (A mode) having an asymmetric displacement. Innumerable high-order n-order modes (Sn, An) exist for each of the fundamental wave symmetric mode (S ₀ ) and the fundamental wave asymmetric mode (A ₀ ), and the wave becomes complicated.
In our study, the oscillation broadband Lamb wave, by geophone, symmetric / method for separating an asymmetric mode is configured, the results of the analysis for each mode by utilizing this, S ₁ mode delamination section S ₀ mode and is converted to a ₁ mode propagates, was found to again propagate back to S ₁ mode passes through the delamination portion at.
Further, A ₁ mode is converted into the fast S ₀ mode than A ₁ mode of propagation speed and propagation in delamination portion was found to propagate back again A ₁ mode passes through the delamination portion.
That is, it has been found that the speed change at the delamination part changes the arrival time, and the arrival time of each mode shows its own change depending on the length of the vibration propagation direction of the delamination part.
Therefore, according to the present invention, the propagation intensity distribution data expanded two-dimensionally in frequency and propagation time is obtained, and the predetermined feature value that reveals the change in the arrival time of this mode due to the influence of damage from the data for a specific mode By obtaining (an index of damage scale), there is an effect that it is possible to diagnose the presence and scale of damage.

It is a schematic structure figure of a damage detection system concerning one embodiment of the present invention. FIG. 2 is a schematic configuration diagram (a) of an optical fiber sensor, and a diagram (b) showing a change in refractive index of a grating part in a light traveling direction. FIG. 2 is a configuration diagram (a) of an optical fiber sensor and a spectrum analyzer connected thereto according to an embodiment of the present invention, and a spectrum diagram (b) showing passbands of eight optical filters. FIG. 4 is an input wave waveform (a) for the optical filter according to an embodiment of the present invention (a), a spectrum diagram (b) showing the pass band of the two optical filters, and an output wave waveform (c) of the optical filter. It is a block diagram which shows the control system of the damage detection system which concerns on one Embodiment of this invention. It is the input waveform (a) to the MFC actuator concerning an experiment, and its Fourier spectrum (b). The received waveform (a) received by the FBG sensor in the experiment, the Fourier spectrum (b), and the wavelet transform result (c). It is a theoretical dispersion curve of Lamb waves under the same conditions as the experiment. It is a conceptual diagram of the mode separation method using an MFC actuator. It is a conceptual diagram of the mode separation method using an FBG sensor. It is a figure which shows the mode identification result which concerns on experiment. Conceptual diagram of Lamb wave mode conversion behavior (a), theoretical dispersion curve (b) of Lamb wave propagating through a plate with a thickness of 3.4 (mm), and propagation through a plate with a thickness of 1.7 (mm) This is the theoretical dispersion curve (c) of the Lamb wave. It is sectional drawing of the test piece which concerns on experiment. It is a figure which shows the mode conversion behavior of S mode calculated | required from experiment. It is a figure which shows the mode conversion behavior of A mode calculated | required from experiment. It is sectional drawing of a finite element analysis model. It is a figure which shows the mode conversion behavior of S mode calculated | required from the finite element analysis. It is a figure which shows the mode conversion behavior of A mode calculated | required from the finite element analysis. The difference in propagation form between when the structure is healthy (a) and when peeling occurs (b) is shown. It shows the speed difference between the S ₀ mode at the speed the release of the A ₁ mode in the healthy section. It is a perspective view of the measurement object part which concerns on the detection experiment of artificial delamination. It is a plot of time wavelet coefficient maximum value of A ₁ mode occurs for each frequency for peeling different length each specimen. For each frequency for peeling different length each specimen is a plot of S _0, S ₁ mode wavelet coefficients times the maximum value occurs for. It is the graph which showed the change with respect to the peeling length of the inclination of the approximate straight line of the measurement data group of each test piece in the range of 250-400 (kHz) of FIG. It is a graph showing a change with respect to the amount of decrease in peel length of the propagation time of the A ₁ mode in 300kHz. 350kHz is a graph showing a change with respect to (finite element analysis) and 400kHz (experiment) in the S _0, S ₁ mode propagation increase in peel length of time.

  An embodiment of the present invention will be described below with reference to the drawings. The following is one embodiment of the present invention and does not limit the present invention.

[Basic configuration]
First, the basic configuration of the damage detection system of the present embodiment will be described.
FIG. 1 is a schematic configuration diagram of a damage detection system 10 that detects damage of a structural composite material Z. In this embodiment, a structural composite material is used as a subject.
In the present embodiment, an MFC (Macro Fiber Composite) actuator is used as a vibration device that applies Lamb wave ultrasonic vibration to a subject. The MFC actuator is a structure in which ultra-thin prismatic piezoelectric ceramics are arranged in one direction and embedded in an epoxy resin, and electrodes are bonded to the upper and lower surfaces thereof, and a relatively large strain can be generated in one direction in the surface. It is known from its characteristics that it can be used as an ultrasonic oscillator. Another oscillation actuator such as a piezo element may be applied as the vibration device.

  As shown in FIG. 1, the damage detection system 10 of this embodiment includes an MFC actuator 21 attached to a surface portion of a structural composite material Z in the vicinity of a location where damage detection of the structural composite material Z is to be performed, The wavelength characteristics of the reflected light obtained from the optical fiber sensor 30 as a vibration detection sensor, the control device 41 of the MFC actuator 21, and the optical fiber sensor 30 installed in the vicinity of the location where the damage detection of the composite material Z is to be performed. A spectrum analyzer 42 to detect and an arithmetic processing unit 50 that performs arithmetic processing on the output value of the spectrum analyzer 42 are provided. The power supply device 43 of the spectrum analyzer 42 is illustrated.

  When a drive voltage is applied from the outside, the MFC actuator 21 generates a relatively large strain in one direction within the surface. Using this property, the control device 41 applies a driving voltage to the MFC actuator 21 and applies instantaneous vibration to the structural composite material Z.

The optical fiber sensor 30 is an FBG (Fiber Bragg Grating) optical fiber sensor, and has an optical fiber having a grating portion 33 that reflects light of a predetermined wavelength on a core portion 32 as shown in the schematic configuration diagram of FIG. 34.
The optical fiber 34 is connected to a spectrum analyzer 42 at one end thereof, and irradiated light covering a wavelength band in a predetermined range is incident on the core unit 32 by a light source included in the spectrum analyzer 42. The light incident from the spectrum analyzer 42 propagates through the core portion 32 and only a part of the wavelength light is reflected by the grating portion 33.

FIG. 2B is a diagram showing the refractive index change in the light traveling direction of the core portion 32, and a range L in the drawing indicates the refractive index in the grating portion 33.
As shown in the figure, the grating portion 33 is configured to change the refractive index of the core portion 32 at a constant period. The grating 33 selectively reflects only light of a specific wavelength at the boundary where the refractive index changes. When a disturbance such as strain is applied to the grating section 33 by vibration, the wavelength of the reflected light changes with the change (expansion / contraction) of the lattice spacing.
Here, the wavelength change Δλ _B of the reflected light of the FBG optical fiber sensor is the effective refractive index of the core n, the grating interval Λ, the Pockels coefficient P ₁₁ , P ₁₂ , the Poisson ratio ν, the applied strain ε, the fiber When the temperature coefficient of the material is α and the temperature change is ΔT, it is expressed by the following equation (1) (Alan D. Kersey, Fiber Grating Sensors “JOURNAL OF LIGHTWAVE TECHNOLOGY, Vol. 15, No. 8, 1997).

Therefore, when vibration occurs in the grating section 33, the distortion amount ε of the grating section 33 changes, and as a result, the wavelength of reflected light varies according to the distortion amount ε. If the vibration is transmitted from the vibration source satisfactorily, the grating section 33 is greatly distorted, and the wavelength change amount Δλ _B greatly fluctuates. If the vibration is not transmitted well from the vibration source, The grating section 33 is slightly distorted, and the wavelength change amount Δλ _B is small and fluctuates.

  The MFC generates a vertical strain in the axial direction of the fibrous piezoelectric element, and the FBG detects the axial strain generated in the fibrous optical fiber. These elements have a wide frequency characteristic without having a resonance frequency, and their behavior has a strong directivity, so that the propagation path is clear. With these two features, this measurement system can propagate wideband Lamb waves with directivity. In addition, FBG and MFC are both small and light, flexible and have high strain at break, so they can be integrated with laminates and are highly reliable without breaking even under large strain. It also has suitable characteristics.

FIG. 3A shows a configuration example of the optical fiber sensor and the spectrum analyzer 42 connected thereto. As shown in FIG. 3A, the spectrum analyzer 42 includes a light source 61, an optical circulator 62, an AWG module 63, and a photoelectric converter 60. In this configuration example, an optical fiber 34 in which four optical fiber sensors 30 a to 30 d having different reflection wavelengths are provided in series is connected to a spectrum analyzer 42. As a minimum configuration, only one optical fiber sensor 30 is required.

The light source 61 is a broadband light source that includes the entire vibration range of the reflection wavelength of the optical fiber sensors 30a to 30d. This is because even when the reflection wavelength of the optical fiber sensors 30a to 30d vibrates due to the Lamb wave, complete reflected light is always obtained.
The optical circulator 62 advances the light from the light source 61 toward the optical fiber sensors 30 a to 30 d, and guides the reflected light from the returned optical fiber sensors 30 a to 30 d to the optical fiber 69. The reflected light led to the optical fiber 69 is introduced into the input port P0 of the AWG module 63.

  The AWG module 63 has an AWG substrate 64. On the AWG substrate 64, an optical wave circuit monolithically integrated on a glass substrate is formed by an optical waveguide technique. The lightwave circuit on the AWG substrate 64 includes input / output slab waveguides 65 and 66, an arrayed waveguide 67, and an output waveguide 68, and includes eight optical filters having different passbands connected in parallel to the input port P0. It is composed. The light wave circuit on the AWG substrate 64 distributes the wavelength-multiplexed input light and passes it through eight optical filters to separate the wavelengths, and outputs them in parallel to the eight output ports P1 to P8. However, the number of output ports in practical use is not limited to eight.

The passband of each optical filter corresponding to the eight output ports P1 to P8 is shown in the spectrum diagram of FIG. For example, in FIG. 3 (b), an optical fiber sensor 3 having a reflection wavelength of the center wavelength λ2.
Reflected light corresponding to the portion where the reflected light input distribution 70 of 0b overlaps the passband 71 is passed through one optical filter and output to the output port P3, and at the same time, corresponds to the portion overlapping the passband 72. The reflected light passes through another optical filter and is output to the output port P4. Similarly, the output ports P1 and P2 are provided for the optical fiber sensor 30a having the reflection center wavelength λ1, the output ports P5 and P6 are provided for the optical fiber sensor 30c having the reflection center wavelength λ3, and the optical fiber sensor 30d having the reflection center wavelength λ4. The output ports P7 and P8 correspond to each other, and wavelength separation is possible by the same principle. As described above, the optical fiber sensor 30 may be one as the minimum configuration, and in this case, two optical filters are sufficient.

As a representative, the processing content for the reflected light from one optical fiber sensor 30 will be described with reference to FIG.
As shown in FIG. 4B, an input distribution 73T of reflected light from the optical fiber sensor 30 appears. During excitation by the MFC actuator 21, a Lamb wave using the MFC actuator 21 as an oscillation source propagates through the structural composite material Z, and the optical fiber sensor 30 outputs a reflection according to the Lamb wave transmitted from the structural composite material Z. Vibrate the wavelength of light. If the vibration of this wavelength is illustrated, it will become the input wave 73W of Fig.4 (a).
Due to the vibration of this wavelength, the reflected light input distribution 73T shown in FIG. 4 (b) alternately shifts upward and downward and vibrates slightly, and the value of the wavelength repeatedly increases and decreases.
In such a wavelength vibration, 73C in the figure is the vibration center of the center wavelength of the reflected light input distribution 73T. The center wavelength 75C of the passband 75T of one optical filter is fixed to the upper region of the vibration center 73C. The center wavelength 74C of the passband 74T of the other optical filter is fixed to the lower region of the vibration center 73C.
The center wavelength 75C and the center wavelength 74C are fixed at positions away from the vibration center 73C by more than the amplitude of the wavelength vibration of the reflected light.
Further, when the reflected light input distribution 73T is stationary, the lower slope 75T-1 of the upper passband 75T intersects the upper slope 73T-1 of the reflected light input distribution 73T, and is reflected from the upper passband 75T. The light input distribution 73T overlaps with a width greater than the amplitude of the wavelength vibration.
Similarly, when the reflected light input distribution 73T is stationary, the upper slope 74T-1 of the lower passband 74T intersects with the lower slope 73T-2 of the reflected light input distribution 73T, and the lower passband 74T and The reflected light input distribution 73T overlaps with a width greater than the amplitude of the wavelength vibration.
By fixing the passband 75T and the passband 74T in the above positional relationship with respect to the reflected light input distribution 73T, the wavelength vibration of the reflected light can be detected with high sensitivity.
The upper optical filter passes and outputs the reflected light corresponding to the portion where the reflected light input distribution 73T overlaps the passband 75T. Similarly, the lower optical filter passes and outputs the reflected light corresponding to the portion where the reflected light input distribution 73T overlaps the passband 74T.

Therefore, when the value of the wavelength of the reflected light increases and the reflected light input distribution 73T shifts upward, the output value of the upper optical filter having the pass band 75T increases, and the output of the lower optical filter having the pass band 74T. The value decreases. Conversely, when the value of the reflected light wavelength decreases and the reflected light input distribution 73T shifts downward, the output value of the upper optical filter having the pass band 75T decreases, and the lower optical filter having the pass band 74T decreases. The output value increases.
Therefore, when the change in the center wavelength of the reflected light is vibrated by the input wave 73W shown in FIG.
The output value of the upper optical filter having the pass band 75T generates the output wave 75W shown in FIG. 4C, and the output value of the lower optical filter having the pass band 74T is the output shown in FIG. 4C. Wave 74
W is generated. As shown in FIG. 4C, the output wave 74W and the output wave 75W are in antiphase waves.

  Based on the above principle, the spectrum analyzer 42 shown in FIG. 3 outputs light waves to each of the eight output ports P1 to P8 during excitation, and these are converted into electrical signals by the photoelectric converter 60 and output externally. The output of the spectrum analyzer 42 is A / D converted via an interface (not shown) and input to the arithmetic processing unit 50.

  As shown in FIG. 5, the arithmetic processing unit 50 includes a CPU 51 that performs arithmetic processing according to a program, a ROM 52 that stores programs for performing various types of processing and control, and temporarily stores data and the like in various types of processing. A RAM 53 serving as a work area, an interface 54 for transmitting / receiving data to / from the control device 41, an interface 55 for inputting data from the spectrum analyzer 42, and display data obtained as a result of calculation into an image signal of a proper format for the display device 56. An image output interface 57 for converting and outputting to the display device 56, and a data bus 58 for transmitting various commands or data between the above components.

  The damage detection system 10 applies vibration to the structural composite material Z by the MFC actuator 21 installed in the structural composite material Z to be damaged, and propagates the vibration wave detected by the optical fiber sensor 30. From this, it is detected whether or not damage has occurred in the vicinity of the optical fiber sensor 30. For this purpose, the arithmetic processing unit 50 executes various functions as described below as the CPU 51 processes various programs stored in the ROM 52 using the RAM 53.

  The CPU 51 controls the operation of the control device 41 so as to apply a driving voltage to the MFC actuator 21 in accordance with a program stored in the ROM 52. When there are a plurality of MFC actuators 21, any MFC actuator 21 may be selected. For example, when a vibration generation source is used, a structural composite is provided between the grating portion 33 of the optical fiber sensor 30. It is desirable to select an MFC actuator that has a portion that is prone to damage to the material Z.

  The CPU 51 applies a driving voltage in accordance with a program stored in the ROM 52, acquires output wave data output in parallel from the spectrum analyzer 42 during a certain period during the excitation by the MFC actuator 21, and stores the output wave data in the RAM 53. I do.

The CPU 51 issues a control command to vibrate the ultrasonic vibration of the Lamb wave to the structural composite material Z by the MFC actuator 21, and the differential signal f between the output wave 74W and the output wave 75W of the optical filter obtained at the time of the vibration. (T) is obtained numerically. For example, the differential signal f (t) shown in FIG.
Then, the CPU 51 performs wavelet transform on the f (t) data by the following equation (2). As a result, the f (t) data is converted into propagation intensity distribution data developed in two dimensions of frequency and propagation time. This data corresponds to the propagation intensity distribution of the Lamb wave to the optical fiber sensor 30, and is as shown in FIG. 7C, for example.

[Damage detection operation]
Using the basic configuration described above, and further installing the MFC actuators 21 and 21 and the optical fiber sensors 30 and 30 at the same position on the front and back of the structural composite material Z as shown in FIG. 19 or FIG. Perform the action.

The CPU 51 causes the MFC actuators 21 and 21 on the front and back to oscillate in-phase waves, so that only the symmetric mode is vibrated to the structural composite material Z, and the wavelet transform is applied to the f (t) data as described above. For example, as shown in FIG. 11A, two-dimensional development data of the frequency and propagation time of only the S mode is obtained. Then, CPU 51 is calculated based on the theoretical dispersion curve illustrated in FIG. 8, S ₀ mode, S ₁ mode, to identify the mode such as S ₂ mode, the propagation time is the maximum wavelet coefficient value for each frequency of a particular mode occurs To do. For example, when a particular mode S ₀ mode and S ₁ mode, the frequency of the maximum wavelet coefficient value occurs as illustrated in FIG. 23, the relationship between the propagation time is specified. This is one feature value extracted from the two-dimensional development data and one measurement result.
The CPU 51 displays this on the display device 56 as shown in FIG. The CPU 51 displays the measurement result for the structure having the same damage state and the same measurement result for the subject whose damage state is unknown. The inspector can refer to this, and the presence or absence and scale of damage can be estimated by comparing them.
Alternatively, since the propagation times of the S ₀ mode and the S ₁ mode increase with an increase in the delamination length, the CPU 51 displays the propagation time in the form of an increase in the case of no damage as shown in FIG. 26, for example. . The examiner can refer to this to estimate the presence or absence of damage and the scale.
Further, the CPU 51 further proceeds to perform an estimation calculation of the damage scale for the subject based on the measurement result for the structure having a known damage state accumulated in the ROM 52 and the like and the measurement result for the subject whose damage state is unknown. The result may be displayed on the display device 56.
In order to acquire the data of the symmetric mode (S mode), instead of the above method, the output values of the optical fiber sensors 30 and 30 on the front and back sides are added to cancel the asymmetric mode and emphasize the symmetric mode. Two-dimensional development data (propagation intensity distribution data) may be obtained.

Further, the CPU 51 oscillates waves of opposite phases in the front and back MFC actuators 21 and 21 to vibrate only the asymmetric mode to the structural composite material Z, and performs wavelet transform on the f (t) data as described above. Thus, for example, as shown in FIG. 11B, two-dimensional development data of the frequency and propagation time of only the A mode is obtained. Then, CPU 51, based on the theoretical dispersion curve illustrated in FIG. 8, A ₀ mode, to identify the mode, such as A ₁ mode, calculates the propagation time is the maximum wavelet coefficient value for each frequency of a particular mode occurs. For example, when a particular mode A ₁ mode, the frequency of the maximum wavelet coefficient value occurs as illustrated in FIG. 22, the relationship between the propagation time is specified. This is one feature value extracted from the two-dimensional development data and one measurement result.
The CPU 51 displays this on the display device 56 as shown in FIG. The CPU 51 displays the measurement result for the structure having the same damage state and the same measurement result for the subject whose damage state is unknown. The inspector can refer to this, and the presence or absence and scale of damage can be estimated by comparing them.
Or, due to the reduced conversion to A ₁ mode propagation time fast S ₀ mode than A ₁ mode of propagation velocity of the damaged portion of, CPU 51 is the propagation time for the absence of injury, as shown in FIG. 25 for example Display in the form of a decrease. The examiner can refer to this to estimate the presence or absence of damage and the scale.
The CPU51 relates _{A 1} mode, to calculate the rate of change of frequency versus propagation time. For example, an approximate straight line is calculated for the measurement data group of each test piece in the range of 250 to 400 (kHz) in FIG. This is also one feature value extracted from the two-dimensional development data and one measurement result. The CPU 51 displays the rate of change (slope) as a numerical value or a graph as shown in FIG. 24, for example. Also here, the CPU 51 displays the measurement result for the structure having the same damage state and the same measurement result for the subject whose damage state is unknown. The inspector can refer to this, and the presence or absence and scale of damage can be estimated by comparing them.
Further, the CPU 51 further proceeds to perform an estimation calculation of the damage scale for the subject based on the measurement result for the structure having a known damage state accumulated in the ROM 52 and the like and the measurement result for the subject whose damage state is unknown. The result may be displayed on the display device 56. The basic data of the estimation calculation is any one or more of the increase amount of the propagation time of the S ₀ mode and the S ₁ mode, the decrease amount of the propagation time of the A ₁ mode, and the change rate (slope). .
In order to acquire data of the asymmetric mode (A mode), instead of the above method, the symmetrical mode is canceled by subtracting the output values of the front and back optical fiber sensors 30 and 30 to emphasize the asymmetric mode. Two-dimensional development data (propagation intensity distribution data) may be obtained.

  In the above embodiment, the differential signal of the output values of the two optical filters is used as the basic data of the wavelet transform. However, the present invention is not limited to this, and the output value of one optical filter is converted to the wavelet transform. It may be basic data.

  In the above embodiment, the maximum peak value of the wavelet coefficient in the specific mode is calculated. However, any parameter value may be used as long as the parameter is suitable for the comparison of the specific mode of the acquired Lamb wave.

  Further, in the above embodiment, the wavelet transform is applied as a transform that expands the detection value of the optical fiber sensor in two dimensions of frequency and propagation time. However, the present invention is not limited to this, and short-time Fourier transform or the like. Other conversion methods may be applied.

[Verification experiment and analysis]
Next, for the purpose of providing a theoretical explanation of the present invention and a reference for carrying out the present invention, the following verification contents by experiment and analysis will be disclosed.

[1. Mode identification method (mode separation method)
First, measurement was performed on a quasi-isotropic CFRP laminate (T700S / 2500, Toray Industries, [45/0 / -45 / 90] 3s, thickness 3.4 mm). MFC (M-2814-P2, Smart material Co., Ltd.) is 6mm long, 14mm wide, 0.3mm thick, FBG sensor (Fujikura Co., Ltd.) is 1.5mm long, with polyimide coating and 150μm diameter. is there. Both were separated from each other by 100 mm and pasted on the surface of the CFRP laminate, and measurement was performed. For attachment, Aron Alpha (Konishi Co., Ltd.), which is a cyanoacrylate adhesive, was used. As an input signal to the MFC, a broadband signal obtained by multiplying one cycle of a sine wave of fc = 400 kHz as shown in FIG. 6 by a Hamming window was used. The received waveform of the Lamb wave oscillated by the MFC, propagated through the laminated plate and received by the FBG sensor was averaged 32768 times in order to remove noise. Thereafter, signal analysis was performed on the obtained received waveform, and the mode dispersion included in the received wave was expressed in the time-frequency domain. For signal analysis, a complex Morlet function was used as a window function, and a one-dimensional complex continuous wavelet analysis was performed. FIG. 7 shows a received waveform received by the FBG sensor, its Fourier spectrum, and a wavelet transform result. From this result, it can be confirmed that a component over a wide band is received without a large peak appearing at a specific frequency. Further, from the wavelet transform result, it is understood that a plurality of modes having different speeds and frequencies are observed and there is mode dispersion. Next, a theoretical dispersion curve is derived in order to identify each mode appearing in this received waveform.

Fig. 8 shows the theoretical dispersion curve of the arrival time at a propagation distance of 100mm in the 3.4mm CFRP laminate as in the above experiment. On the dispersion curve, the arrival time of the higher-order mode is abruptly delayed, and the frequency that diverges to infinity is called the cut-off frequency. Comparing this theoretical dispersion curve with the wavelet transform result of the received waveform, it can be seen that the mode dispersion is in good agreement between the two. However, it is difficult to identify the mode in the frequency region of 300 kHz or more where multiple modes overlap, and in order to perform accurate mode identification,
It is necessary to separate these overlapping modes.
Therefore, as a method for separating these modes, a method of attaching the MFC and the FBG sensor to the same position on the upper and lower surfaces of the laminated plate was used. As shown in FIG. 9, when the MFCs are attached to the same position on the upper and lower surfaces and oscillate waves having the same phase, it is possible to oscillate only the symmetric mode. On the contrary, if an antiphase wave is oscillated, only the asymmetric mode can be oscillated.
In addition, as shown in FIG. 10, the FBG sensor is attached to the same position on the upper and lower surfaces, and the symmetric mode can be separated by taking the sum of the received waveforms on the upper and lower surfaces of the plate, and the asymmetric mode can be separated by taking the difference. I can do it.
Using these two methods, S (symmetric) mode and A (asymmetric) mode included in the received Lamb wave were separated, and after wavelet transform, respectively, mode identification was performed by comparing with the previous theoretical dispersion curve. The results are shown in FIG. As a result, by mode separation, overlapping of multiple modes was eliminated, and mode identification could be performed accurately. Then, the received oscillation waveform _{A 0, S 0, A 1} , S 1, _{that S 2} mode is present is confirmed.
From the above results, each mode included in the received Lamb wave can be identified by using the mode separation method.

[2. Cause of change in propagation time of specific mode (mode conversion behavior at delamination)
In the previous section, each mode was identified, and it became possible to understand the mode dispersion included in the measurement results. Next, the mode conversion behavior resulting from this change in mode dispersion will be elucidated by experiments and analysis.

(1) Mode conversion due to plate thickness change of peeled portion Since the propagation speed of Lamb waves depends on the product of frequency and plate thickness, the mode dispersion of Lamb waves also changes when the plate thickness changes. Therefore, as shown in FIG. 12, when delamination occurs inside the laminated plate, the plate thickness of the propagation path in the peeled portion is smaller than that in the healthy portion, so that the mode dispersibility differs between the healthy portion and the peeled portion. Due to this change in mode dispersibility, mode conversion occurs in the Lamb wave that has propagated through the healthy part, and it is considered that the separation part propagates in a mode mode different from the healthy part.
For example, in a laminated plate with a thickness of 3.4 mm, there are three modes of A ₀ , S ₀ , and A ₁ modes as a propagation mode of a Lamb wave with a frequency of 300 kHz, but delamination occurs at the center of the laminated plate and When the plate thickness changes to 1.7 mm, there are only two propagation modes, A ₀ and S ₀ modes.
Therefore, a healthy section Lamb waves propagating as A ₁ mode is mode converted by peeling section, so that the propagated as A _0, S ₀ mode. However, it cannot be determined from the theoretical dispersion curve which mode propagates through the peeled portion. Therefore, the mode conversion behavior that occurs at the actual delamination is clarified through experiments and finite element analysis.

(2) Experiments In order to elucidate the mode conversion behavior that occurs at the actual delamination start edge, pseudo-isotropic CFRP laminates (T700S / 2500, Toray Industries, Inc., [45/0 / -45 / 90 3s, thickness 3.4mm) was simulated when there was delamination at the center in the thickness direction. In addition, since the mode identification of the received Lamb wave is performed by mode separation, in order to measure the mode dispersibility of the peeling portion, it is necessary to attach an FBG sensor to the inner surface of the simulated delamination. Therefore, two CFRP laminates with a thickness of 1.7mm are prepared, and after bonding the FBG sensor to one side, epoxy-based bonding is applied to the 60mm length range from one end of the plate so that the surface to which the FBG sensor is bonded comes inside. Adhesion was performed with the agent Araldite Standard (Huntsman Advanced Material Co., Ltd.). Two CFRP laminates were made with a laminate configuration [45/0 / -45 / 90] 3 to simulate the laminate configuration [45/0 / -45 / 90] 3s, and are symmetric on the bonding surface Glued together. The dimensions of the test piece are shown in FIG. The width of the plate is 90 mm.
MFC (M-2814-P2) having a length of 6 mm, a width of 14 mm, and a thickness of 0.3 mm was used and attached to the upper and lower surfaces of the laminate one by one. Lamb waves are attached by attaching FBG sensors to the top and bottom surfaces of the laminate at two points: a distance of 30 mm from the MFC tip and a thickness of 3.4 mm (healthy part) and a distance of 70 mm and a thickness of 1.7 mm (peeling part). Receive vibration. The FBG sensor (Fujikura Co., Ltd.) used in the experiment has a sensor length of 1.5 mm, a polyimide coating, and a diameter of 150 μm. Aron Alpha (Konishi Co., Ltd.) was used for pasting the element. The input signal is a signal obtained by multiplying one cycle of a sine wave of fc = 400 kHz by a Hamming window, and the received waveform averages 32768 measurement waveforms in order to remove noise. FIG. 14 shows the mode conversion behavior of the S mode obtained by oscillating only the S (symmetric) mode using the MFCs of the upper and lower surfaces. FIG. 14 shows the A mode obtained by oscillating only the A (asymmetric) mode. The mode conversion behavior is shown in FIG.

From the result of FIG. 14, when only the S mode is oscillated, only the S ₀ mode and the S ₁ mode are observed in the healthy part. Than the stoichiometric dispersion curve for plate thickness 1.7mm shown in FIG. 12 (c), since the peeling section it has been found that there is only the S ₁ mode or 800 kHz, S ₁ mode that has been observed in a healthy portion of the release It is considered that the mode is changed in the part and propagated as another mode. Therefore, when observing the modes that exist in the peeling section, two S ₀ mode and A ₁ mode is observed. From this result, it was confirmed that mode conversion of “S ₁ mode → S ₀ mode, A ₁ mode” occurred at the start of delamination.
From the results of FIG. 15, when only the A mode is oscillated, only the A ₀ mode and the A ₁ mode are observed in the healthy part. Figure 12 than the theoretical dispersion curve in the case of thickness 1.7mm to (c), the order in peeling section has been found that the A ₁ mode only exists in more than 500 kHz, A ₁ mode that has been observed in a healthy portion of the release It is considered that the mode is changed in the part and propagated as another mode. Therefore, when observing the modes that exist in the peeling section, two is observed in S ₀ mode and A ₁ mode (A ₀ mode is not taken into consideration as a target of slow for mode conversion arrival time). The peeling section A ₁ mode that has been observed in not a mode conversion is believed that the above A ₁ mode 500kHz propagated as it is. Therefore, from this result, it was confirmed that mode conversion of “A ₁ mode → S ₀ mode” occurred at the delamination start portion.

(3) Verification by finite element analysis In order to verify the mode conversion behavior obtained from the experiment of (2) above, a two-dimensional finite element analysis was performed. FIG. 16 shows a finite element model and its dimensions. LS-DYNA 971 was used for model construction and analysis. 2Dshell element (plane strain) was used as an element of the analysis model. The mesh size was set to 0.125mm, which was small enough to calculate high frequencies with short wavelengths. The contact portion between the MFC and the CFRP laminate was node-coupled, and the input waveform to the MFC was a sine wave with a frequency of 400 kHz multiplied by a Hamming window as in the experiment. Also, since the piezoelectric effect cannot be calculated with LS-DYNA, the piezoelectric effect was simulated as the coefficient of thermal expansion in the expansion and contraction direction of MFC. Under the conditions as described above, the three receiving points shown in FIG. 16 (sound part: propagation distance 20 mm, plate thickness 3.4 mm, peeled part: propagation distance 60 mm, plate thickness 1.7 mm, sound part (after peeling): At a propagation distance of 100 mm and a thickness of 3.4 mm, the time history of strain in the x direction was calculated and used as the received waveform. FIG. 17 shows the mode conversion behavior of the S mode obtained by oscillating only the S (symmetric) mode using the MFCs of the upper and lower surfaces. FIG. 17 shows the A mode obtained by oscillating only the A (asymmetric) mode. The mode conversion behavior is shown in FIG.

From the results shown in FIG. 17, it was confirmed that mode conversion of “S ₁ mode → S ₀ mode, A ₁ mode” occurred at the start of delamination as in the experiment. Moreover, in the healthy section after delamination passage, similar dispersibility and healthy section before passing the peeling was observed, S ₀ mode and S ₁ mode is observed. In this case S ₁ mode back to healthy section propagates through the peeling section, S ₀ mode peeling section, A ₁ mode, presumably again the mode conversion to the S ₁ mode in the healthy section.
Next, from the result of FIG. 18, it was confirmed that mode conversion of “A ₁ mode → S ₀ mode” occurred at the delamination start portion as in the experiment. Moreover, in the healthy part after delamination passage,
Dispersibility similar to that of the sound part before passing through peeling was observed, and A ₀ mode and A ₁ mode were observed.
The A ₁ mode when returning to healthy section propagates through the peeling section, S ₀ mode peeling section is considered again and mode conversion to the A ₁ mode in the healthy section.
From the above, the validity of the mode conversion behavior obtained by experiments was demonstrated, and the mode conversion behavior at the point where delamination ends was also clarified. As a result, it was confirmed that the following two mode conversion behaviors exist when passing through the delamination part.
- _{"S 1} mode → _{S 0} mode, _{A 1} mode → _{S 1} mode"
- _{"A 1} mode → _{S 0} mode → _{A 1} mode"

Due to such mode conversion behavior, the mode when propagating through the peeled portion is different from the mode when propagating when healthy. For example, in the mode conversion of “A ₁ mode → S ₀ mode → A ₁ mode”, as shown in FIG. 19, all propagation paths propagate as A ₁ mode when healthy, but when separation occurs, the region Are propagated as S ₀ modes. In this case, sound unit (thickness 3.4 mm) propagation velocity of the A ₁ mode and the release portion S ₀ mode (thickness 1.7 mm) are different as shown in FIG. 20, S ₀ mode it is A ₁ Faster than mode. Thus, A ₁ mode arrival time is geophone in FBG sensor is earlier than when healthy person when peeling occurs. This change in arrival time occurs according to the difference in propagation speed between the mode propagating through the healthy part and the mode propagating through the peeling part, and the length of peeling. Therefore, by using this difference as an index, it is possible to detect delamination and quantitatively evaluate the delamination length.

[3. (Verification experiment and analysis by artificial delamination detection)
(1) Verification experiment It is verified from an experiment whether or not a change in arrival time actually occurs in the wave after passing through the peeling portion, and the effectiveness of the present invention is shown.
Therefore, during the formation of CFRP pseudo-isotropic laminates, by embedding two 50μm thick Teflon (registered trademark) films between adjacent 90 ° layers in the center of the plate thickness direction, In addition, three types of laminates were prepared by artificially introducing delamination with a separation length L = 20, 40, 60 mm. And the broadband Lamb wave was propagated so that it might pass through those peeling area | regions, and the received waveform was measured. The experimental configuration is shown in FIG. The MFC and FBG sensors are bonded to the same location on the top and bottom surfaces of the plate, respectively, and mode separation is performed in the same manner as when the mode conversion behavior was clarified in the previous section.
Using this experimental configuration, an artificial delamination detection experiment was performed using a laminate having artificial peeling of L = 20, 40, 60 mm introduced. Compare these results with the results of measurements on sound laminates (L = 0mm), and evaluate changes in arrival time.
For this purpose, the wavelet transform of the received waveform was performed, and then the maximum value of the wavelet coefficient of each frequency was extracted. When there is delamination, the amount of change from the sound state at the time of the maximum value of the wavelet coefficient corresponds to the amount of change in arrival time.
When the A mode is oscillated using the upper and lower surface MFC, the “time for which the wavelet coefficient maximum value appears for each frequency” of the A mode measured by the FBG sensor is L = 0, 20, 40, 60 mm. respect a ₁ mode 200~700kHz relatively large change has occurred in the arrival time, shown in FIG. 22.
Next, when the S mode is oscillated using the upper and lower surface MFCs, the “time when the wavelet coefficient maximum value for each frequency appears” of the S mode measured by the FBG sensor is set to L = 0, 20, 40, 60 mm. FIG. 23 shows the S ₀ and S ₁ modes of 400 to 600 kHz in which the arrival time varies with respect to the case.
From FIG. 22, how the arrival time of the A ₁ mode becomes earlier is observed as the length of the peeling becomes long. It is also observed that the slope of mode dispersion in the frequency region of 200 to 500 kHz changes. This is because, as shown in FIG. 20, the difference between the propagation speeds of the S ₀ mode and the A ₁ mode increases as the cutoff frequency becomes closer to the A ₁ mode (lower frequency).
Further, from FIG. 23, it is observed that the arrival time of the S ₀ and S ₁ modes is delayed as the separation length increases.
From the above results, when the delamination occurs, the arrival time is surely changed, indicating that the present invention is effective.

(2) Verification by the finite element method 2. The delamination length of the two-dimensional finite element analysis model used in (3) was changed at L = 20, 40, 60 mm, and the analysis was performed with the same experimental configuration as the above experiment. After that, as in the experiment, the maximum amplitude is obtained for the A ₁ mode and the S ₀ , S ₁ mode, and when the change in arrival time is observed, the result agrees well with the result of the experiment (FIGS. 22 and 23). Was observed.

(3) Quantitative evaluation of peel length Further, quantitative evaluation of peel length using changes in arrival time and slope of mode dispersion observed from the experimental result of (1) and the analysis result of (2) above as indices. The possibility of
A linear approximation line was calculated from a plot of the maximum amplitude value in the frequency domain where the arrival time changed, and the following index was obtained using the approximation line. The index was obtained from the experimental results and the analysis results. The results of plotting each index for each peeling length are shown in FIG. 24, FIG. 25, and FIG.
Figure 24 is a _{A 1} Mode Mode dispersion slope of 250～400KHz, FIG 25 _S is the decrease of _{A 1} mode arrival time at 300kHz, in FIG. 26 is 400kHz (analysis 350 kHz) _{0, S 1} Indicates the increase in mode arrival time.
From the results shown in FIG. 24, it is observed that the slope of the mode dispersibility decreases as the peeling length increases. Further, FIG. 25, from the results of FIG. 26, as the peeling length becomes larger, A ₁ mode and decrease in the arrival time of, S _0, S ₁ increase in the arrival time of the mode how becomes larger observation Is done. Moreover, these indices change almost in proportion to the peel length. Therefore, it is possible to quantitatively evaluate the peeling length by using these indicators.

[4. Conclusion)
As described above, first, each mode of the broadband Lamb wave measured by the broadband ultrasonic transmission / reception system was identified. For this purpose, we proposed a method to separate symmetric / asymmetric modes, and showed that mode identification is possible using this separation method.
Next, the mode conversion behavior in the delamination part is clarified from experiments and analysis, and “S ₁ mode → S ₀ mode, A ₁ mode → S ₁ mode”, “A ₁ mode → S ₀ mode → A ₁ mode” The existence of two mode conversion behaviors was confirmed.
Then, the effectiveness of the peeling detection method according to the present invention using the Lamb wave velocity change by mode conversion was verified from experiments and analyses. As a result, it was confirmed that the speed change in the peeled portion was observed as a change in arrival time.
Finally, the slope of the mode dispersion of the A ₁ mode, the amount of decrease in the arrival time of the A ₁ mode, by each of the three indicators of increasing amounts of S _0, S ₁ mode arrival time, quantitative evaluation of the peel length Showed that it is possible.

DESCRIPTION OF SYMBOLS 10 Damage detection system 21 MFC actuator 30 Optical fiber sensor 32 Core part 33 Grating part 42 Spectrum analyzer 50 Arithmetic processing unit 63 AWG module 64 AWG board | substrates 65 and 66 Input / output slab waveguide 67 Array waveguide 68 Output waveguide P0 Input port P1 ~ P8 Output port 70, 73T Reflected light input distribution
71, 72, 74T, 75T Passing area 73C Vibration center 101 Separation 102 Lamb wave 103 Mode conversion behavior 104 Reflection 105 Adhesive layer

Claims (8)

An excitation device for applying ultrasonic vibration of a broadband Lamb wave to the subject;
A vibration detection sensor for detecting a broadband Lamb wave transmitted from the subject;
A processing device that controls the oscillation operation of the vibration exciting device, processes the output value of the vibration detection sensor, and outputs a measurement result;
The processor is
By converting the output value of the vibration detection sensor obtained at the time of vibration by the vibration device to obtain the propagation intensity distribution data developed in two dimensions of frequency and propagation time, the asymmetric mode of Lamb wave is identified. For one or more specific asymmetric modes selected from the identified Lamb wave asymmetric modes, the propagation time at which the maximum value of the propagation intensity occurs for each frequency is calculated, and the frequency and propagation at which the maximum value occurs. Identify time relationships,
In contrast to the relationship between the identified frequency and propagation time, and the relationship between the frequency and propagation time specified in advance in a subject having a known damage state, which is stored in the storage unit, the damage scale for the subject is estimated. A damage diagnosis system characterized by performing calculations and displaying. The excitation device and the vibration detection sensor are respectively installed at the same position on the front and back of the subject,
The processing device oscillates asymmetrically by oscillating anti-phase waves in the vibration devices on the front and back sides, converts the calculated data by adding the output values of the vibration detection sensors on the front and back sides, and converts the frequency and The damage diagnosis system according to claim 1, wherein the propagation intensity distribution data developed in two dimensions of propagation time is obtained. The processor is
Instead of specifying the relationship between the frequency and propagation time at which the maximum value of the propagation intensity distribution data occurs,
Calculate the propagation time at which the maximum value of the propagation intensity occurs for each frequency and plot the data indicating the relationship between the specified frequency and the propagation time on a graph in which the horizontal axis represents the frequency and the vertical axis represents the propagation time. Calculate the slope of the data in the frequency range as the rate of change,
Comparing the calculated rate of change with the rate of change calculated in advance in a subject whose damage state is known, stored in the storage means, and performing an estimation calculation of the damage scale for the subject The damage diagnosis system according to claim 1, wherein the damage diagnosis system is a damage diagnosis system. The processor is
Instead of calculating the rate of change,
Calculate the amount of decrease in the propagation time at the maximum value of the propagation intensity distribution data with respect to the case where the subject is not damaged,
The damage diagnosis system according to claim 3, wherein an estimation calculation of a damage scale for the subject is performed and displayed based on the calculated amount of decrease. An excitation device for applying ultrasonic vibration of a broadband Lamb wave to the subject;
A vibration detection sensor for detecting a broadband Lamb wave transmitted from the subject;
A processing device that controls the oscillation operation of the vibration exciting device, processes the output value of the vibration detection sensor, and outputs a measurement result;
The processor is
By converting the output value of the vibration detection sensor obtained at the time of vibration by the vibration device, the propagation intensity distribution data expanded in two dimensions of frequency and propagation time is obtained, and the symmetry mode of the Lamb wave is identified. For one or more specific symmetric modes selected from the symmetric modes of the identified Lamb wave, the propagation time at which the maximum value of propagation intensity occurs for each frequency is calculated, and the frequency and propagation at which the maximum value occurs. Identify time relationships,
In contrast to the relationship between the identified frequency and propagation time, and the relationship between the frequency and propagation time specified in advance in a subject having a known damage state, which is stored in the storage unit, the damage scale for the subject is estimated. A damage diagnosis system characterized by performing calculations and displaying. The excitation device and the vibration detection sensor are respectively installed at the same position on the front and back of the subject,
The processing device oscillates symmetrically by causing the vibration devices on the front and back to oscillate in-phase waves, converts the calculated data by adding the output values of the vibration detection sensors on the front and back, and converts the frequency and 6. The damage diagnosis system according to claim 5, wherein propagation intensity distribution data developed in two dimensions of propagation time is obtained. The processor is
Instead of specifying the relationship between the frequency and propagation time at which the maximum value of the propagation intensity distribution data occurs,
Calculating the amount of increase in the propagation time at the maximum value of the propagation intensity distribution data relative to the case where the subject is not damaged;
The damage diagnosis system according to claim 5 or 6, wherein the damage diagnosis system according to claim 5 or 6 displays the estimated damage scale for the subject based on the calculated increase amount.   The processing device displays the relationship between the frequency and the propagation time in the specified asymmetric mode or the symmetric mode, and the calculated change rate of the propagation time, instead of displaying the damage scale for the subject obtained by performing the estimation calculation. 8. The damage diagnosis system according to any one of claims 1 to 7, wherein the calculated decrease amount or the calculated increase amount is displayed.